# Your first discovery project

> Let the Business Analyst build a living model of a project from notes, Slack threads, and meeting recordings, then propose tickets when the picture is complete.

_CommerceCopilot documentation · Getting started_

Discovery is for projects that are not fully specified yet. Instead of forcing you to write a brief before anything can happen, the Business Analyst listens to the places where requirements actually form (notes, Slack conversations, client calls) and maintains a living picture of the project: what the outcome is, what is known, what is assumed, and what still needs an answer.

## Create the project

From the **Projects** page, click **New Project** and switch the toggle to **Start with discovery**.

![The New Project wizard in discovery mode with a project name field and an explanation of how discovery works](/docs/new-project-discovery.png)

Name the project and click **Create & start discovery**. Only the Business Analyst is active at this point. The Tech Lead and Developer agents stay paused until you decide discovery is sufficient, so nothing downstream runs before the requirements are ready.

## Prime the discovery

Right after creation, CommerceCopilot asks you to describe the project in a sentence or two. This is the single highest-leverage minute of the whole flow: a good prime gives the Business Analyst the frame it needs to interpret everything that follows.

A useful prime covers the goal, the main constraint, the deadline, and the ecommerce platform you are building on:

> Aurora Skincare wants to lift mobile conversion on their Shopify Plus store. We are reworking the product detail page, simplifying navigation, and adding a routine-builder quiz. Checkout must keep Shopify Payments. Launch target is late September.

Naming the platform lets the Business Analyst build the plan to its conventions. Leave it out and it will not assume one: it infers the platform from what discovery picks up, and until it knows, it raises "Which ecommerce platform will this storefront be built on?" as an open question rather than defaulting to a guess.

You can skip this step, but expect the first synthesis to ask more open questions in return.

## The discovery dashboard

The Business Analyst runs its first synthesis pass immediately. The project's **Discovery** tab is a live dashboard of everything it has worked out so far:

![The discovery dashboard showing the delivery-readiness gauge, the outcome, the open-question count, the proposed ticket graph, and the discovery to-do list](/docs/discovery-workspace.png)

Each card tracks one part of the living model:

- **Delivery readiness** is a gauge of how much of the proposed work you have accepted for delivery. It climbs as the Business Analyst proposes tickets and you accept them.
- **Outcome** is the Business Analyst's current understanding of what this project is solving for.
- **Open questions** counts the gaps, split between questions tied to specific tickets and project-wide ones. This is the defining behavior of the agent: when something is unclear, it asks instead of inventing an answer.
- **Discovery to-do** is the short list of what needs you next: open questions to answer, assumptions to validate, opportunities to weigh.
- **Delivery tickets** is a thumbnail of the proposed ticket graph. Click it to open the full board.
- **Opportunities, Assumptions, Decisions** capture what could be done, what is being taken on faith, and what has been settled.
- **Data sources** shows which discovery inputs are live for this project (Slack, meetings, your ticket tracker) and links to Settings to change them.

The dashboard refreshes on its own after every new piece of input, and flags what the latest synthesis pass added or changed with a **New** badge. You never need to reload or re-run anything by hand.

## Feed it signal

Discovery gets smarter with everything you give it. Open the ticket board (the **Delivery tickets** card, or the **Tickets** tab) and use the input rail down its left edge. Four kinds of signal feed the model:

### Notes

Open the **Add note** panel from the input rail. Type or paste anything: call notes, a client email, a half-formed thought on scope. The Business Analyst refines the model automatically with every note you add.

![The Add note panel with a free-text field for capturing what you learned](/docs/discovery-add-note.png)

### Slack threads

Once [Slack is connected](/docs/slack), you can bind a channel to the project from its discovery settings. The Business Analyst summarizes the conversation into discovery as it happens, so decisions made in Slack stop getting lost. It posts a one-time disclosure message in the channel before it starts reading, so your client or team knows it is there.

### Meeting calls

The [meeting notetaker](/docs/meeting-notetaker) joins your Zoom, Google Meet, or Teams call and records with the host's permission. The transcript streams into discovery while the meeting is still running: the model refreshes about every five minutes during the call, and the final minutes land shortly after it ends. If you want the Business Analyst to raise its questions live (in the call's chat, spoken aloud, or in a Slack thread), turn on live feedback in the project's discovery settings; the [meeting notetaker guide](/docs/meeting-notetaker) covers it.

### In-person meetings

For a conversation in the room rather than on a call, open the **In-person notetaker** panel, confirm that everyone consents to being recorded, and click **Start recording**. It captures the meeting through your device's microphone, transcribes it as you go, and refreshes the model about every five minutes, so ticket proposals update while you talk. The audio is deleted once it has been transcribed. Both notetakers are covered in [the meeting notetaker guide](/docs/meeting-notetaker).

## From discovery to tickets

As the model fills in, the Business Analyst starts proposing delivery tickets: real, structured work items with business requirements and user stories, visible in the **Delivery tickets** card on the dashboard and on the project's **Tickets** tab. Accept the ones you want delivered, dismiss the rest, and chat on any ticket to refine it. The **Delivery readiness** gauge tracks your progress. When a later pass decides a proposed ticket no longer fits, it flags the ticket for removal rather than deleting it, so you can keep it with one click; the agent never removes work on its own. Development stays gated while discovery is active, and the project header shows the current gate status.

When you judge discovery sufficient, change the project's discovery status from the header dropdown, and enable the downstream agents from the project's **Agents** settings. If a ticket tracker is connected, proposed tickets flow into [Notion](/docs/notion) or [Jira](/docs/jira) where your team already works.

> **Cost note:** discovery consumes credits as it synthesizes. An active discovery project uses roughly 20 credits per month of continuous listening, so the included 135 credits comfortably covers several parallel discoveries.

## Next steps

- [From brief to tickets](/docs/your-first-brief-to-ticket-project), for when the requirements already exist on paper
- [Connect Slack](/docs/slack)
- [Set up the meeting notetaker](/docs/meeting-notetaker)

---
Canonical page: https://www.commercecopilot.ai/docs/your-first-discovery-project